Well Larry, you had to make me do it, didn't you! Dig out the old napkin again, that is... :-). I decided to cook up some parametric spacecraft, based closely on the Gemini, but with modern materials and so forth. You may wish to skip the methodology and jump straight to the conclusion, because this is all rather dense.

=====================
Methodology:

The structural mass of the Gemini re-entry module (638 kg) supported a total habitable volume of 2.55 m^3, which was incredibly small for two people. Any viable tourist craft will need to be a bit more comfortable, so let's call for 1.5 m^3/person for the baseline design. The structural weight should scale more with the surface area than the volume, which means that larger vehicles will be naturally more efficient. Assuming a spherical geometry (wrong, I know, but good enough for the napkin), Gemini had a surface area of 9 m^2, which gives a structural weight of 71kg/m^2 of surface. Let's assume we can use modern materials to better that by 30%, which yields a structural mass for the habitable volume of 50 kg/m^2. On the Gemini, an additional 310 kg of structural mass was required for 1658kg of non-crew-related mass; let's increase the efficiency by 30% again, and that gives us .13 kg per kg of non-crew-related mass. A sloppy way of divvying up the numbers, I'm sure, but probably correct within an order of magnitude.

For the crew mass, Gemini used 498 kg/person (including the person, seat, and provisions). I'm sure the seats have gotten lighter, and since 2-week missions probably aren't called for, the provisions can also be lighter. But the people will usually be larger than Gus Grissom, so let's keep the overall mass budget the same. ECS (117kg) has gotten a lot lighter, and let's distribute it on a per-person basis, adding another 30 kg/person, and let's factor in our misc. mass (175kg) at 30 kg/person, bringing us up to 558 kg/person for the crew. I think this is actually quite generous.

Nav, Telemetry, & Coms (180 kg combined) can probably be cut by 90%. Electrical equipment (420 kg) can probably be cut by 60%. Both should be roughly insensitive to the size of the crew, so let's say that we're down to a fixed 186 kg for these things.

So, the crew, volume-mass, nave, telemetry, coms, Electrical, ECS & Misc. mass for the Gemini totaled 1926 kg. Let's base the rest of the equipment mass off that number (let's call it "system mass," just to be confusing).

The Gemini carried a great deal of delta-v for maneuvering, with its RCS systems, main engine, & propellant totaling 624 kg, or 32.4% of the system mass. Although I don't think a private craft would really need so much delta-v, and I'm sure that the efficiency could be increased in any case, let's keep that percentage the same and use it for a rocket-assisted landing (Soyuz style). Let's also keep TPS & Recovery systems the same at 7.5% and 5.1% of the system mass respectively, again trading modern efficiencies against greater operability and/or comfort. So, all these things (let's call them "recovery systems") now account for 45% of the system mass.

=====================
Conclusion:

Turns out that I think you're pretty much right. Here's what the napkin says:

3-PERSON VEHICLE:
Crew & Support: 1674 kg
Volume & volume structure: 4.5 m^3 / 659 kg
Nav, Telemetry, Elec & Coms: 186 kg
Recovery systems: 1134 kg
Other structure: 172 kg
TOTAL MASS: 3825 kg

...Easy enough to do on a Falcon V. So, how about:

4-PERSON VEHICLE:
Crew & Support: 2232 kg
Volume & volume structure: 6 m^3 / 798 kg
Nav, Telemetry, Elec & Coms: 186 kg
Recovery systems: 1447 kg
Other structure: 212 kg
TOTAL MASS: 4875 kg

...And whoops, we're over the mass budget. Rats. However it's worth noting that SpaceX is right now intentionally downgrading the payload numbers for the Falcon 5, until they can put the actual hardware through its paces to see how it works. They did the same thing for the Falcon 1, and its payload increased 25% in the translation between the drawing board and the test stand, so it's probably realistic to expect the Falcon V to launch a 4-person vehicle. With an equatorial launch site and an equatorial orbit, a 5-person vehicle might even be possible, but like you say, it'd be a stretch.

As for the price per seat, I suspect it'll be somewhat lower than you estimate. As a 5-engine vehicle, the Falcon V has automatic engine-out capability, which I suspect makes it intrinsically more "man-rated" than any vehicle currently in the US fleet. So I don't think that part will add much cost at all. As for the operational cost of the orbiter, I can imagine that being in the neighborhood of $3M per flight. So that gives us a per-seat price of around $5M.

Of course this still isn't great, but at least it would give us a second data point on the price-demand curve for the space tourism market.

But wait, there's more! According to this Aviation Leak article, an upgraded Falcon V is planned to debut in 2007. This will go for $20 million, and feature an RL-10 based 2nd stage, able to put 9000kg (downrated number, presumably) into LEO. Here's what the napkin has to say about that:

9-PERSON VEHICLE:
Crew & Support: 5022 kg
Volume & volume structure: 13.5 m^3 / 1371 kg
Nav, Telemetry, Elec & Coms: 186 kg
Recovery systems: 2961 kg
Other structure: 409 kg
TOTAL MASS: 9949 kg

Assuming an operational cost of $5M/flight for the orbiter, we're now down to $3.2M per seat. I still won't be able to afford a flight, but I take heart when Elon Musk says something like this: "SpaceX plans call for a long term heavy lift vehicle development. I should be clear that Falcon V is not it."

Once the economies of scale get big enough, I expect that I *will* be able to buy a seat. It looks like that won't be happening for a while yet, but I think we *are* heading in the right direction.

Cheers,

Nathan.